. But the sense-datum which we call hearing the thunder does not take
place until the disturbance of the air has travelled as far as to where
we are. Similarly, it takes about eight minutes for the sun's light
to reach us; thus, when we see the sun we are seeing the sun of eight
minutes ago. So far as our sense-data afford evidence as to the physical
sun they afford evidence as to the physical sun of eight minutes ago; if
the physical sun had ceased to exist within the last eight minutes, that
would make no difference to the sense-data which we call 'seeing
the sun'. This affords a fresh illustration of the necessity of
distinguishing between sense-data and physical objects.

What we have found as regards space is much the same as what we find
in relation to the correspondence of the sense-data with their
physical counterparts. If one object looks blue and another red, we may
reasonably presume that there is some corresponding difference between
the physical objects; if two objects both look blue, we may presume a
corresponding similarity. But we cannot hope to be acquainted directly
with the quality in the physical object which makes it look blue or red.
Science tells us that this quality is a certain sort of wave-motion, and
this sounds familiar, because we think of wave-motions in the space we
see. But the wave-motions must really be in physical space, with which
we have no direct acquaintance; thus the real wave-motions have not that
familiarity which we might have supposed them to have. And what holds
for colours is closely similar to what holds for other sense-data. Thus
we find that, although the _relations_ of physical objects have all
sorts of knowable properties, derived from their correspondence with the
relations of sense-data, the physical objects themselves remain unknown
in their intrinsic nature, so far at least as can be discovered by means
of the senses. The question remains whether there is any other method of
discovering the intrinsic nature of physical objects.

The most natural, though not ultimately the most defensible, hypothesis
to adopt in the first instance, at any rate as regards visual
sense-data, would be that, though physical objects cannot, for the
reasons we have been considering, be _exactly_ like sense-data, yet they
may be more or less like. According to this view, physical objects will,
for example, really have colours, and we might, by good luck, see an
object as of the colour it re